一种自适应的白平衡解决方案研究

A Research on an Self-adaptive White Balance Solution Scheme

现有的白平衡算法可以解决大部分的白平衡问题,但没有一种算法能够适用于所有的图像,并且对于低色彩丰富度图像的白平衡校正算法较少。本研究首先针对低色彩丰富度图像提出了一种灰度世界算法结合离散余弦变换的白平衡算法(GW-DCT)。实验证明该算法对于图像中没有单一大色块且色彩丰富度较低的图像的白平衡效果较好,而对以单一颜色为主体的图像效果欠佳。然后提出了一个基于完美反射法(Perfect Reflector Assumption,PR)、正交组合算法(Quadratic Combining GW&PR,QCGP)和GW-DCT的自适应白平衡解决方案,通过图像颜色值的标准差,信息熵,亮度和颜色占比率对图像归类后选择合适的校正算法从而实现白平衡。实验结果表明,该自适应的白平衡解决方案不仅能正确地选择合适的算法,同时对具有不同色彩丰富度的偏色图像都可以获得较理想的白平衡效果。该算法复杂度较低,具有普遍的适应性。

The published white balance algorithms can solve most of the white balance problems,but no one can be suitable for all images.Moreover,there are few white balance correction algorithms for low color richness images.Therefore,an improved white balance algorithm that composites of the gray world and the discrete cosine transform(GW-DCT)is proposed to be appropriate for the low color richness image.The experimental results show that the white balance effect of the algorithm is suitable for the image without single large color block,but not good at the image with that.Then,an adaptive white balance solution based on perfect reflection(PR),Quadratic Combining GW&PR(QCGP)and GW-DCT is proposed in this paper.After classifying images by standard deviation of images color value,information entropy,brightness and color ratio,an appropriate correction algorithm is selected to achieve white balance.The experimental results show that the adaptive white balance solution proposed can not only correctly select the appropriate algorithm,but also obtain ideal white balance effect for color biased images with different color richness.The algorithm has low complexity and universal adaptability.